Discrimination in vitro between the acoustic emissions from Bjork-Shiley convexo-concave valves with and without a broken minor strut.

A statistical pattern recognition technique is used to learn and recognise the frequency spectra of the closing sounds emitted by Bjork-Shiley convexo-concave heart valves, with and without fractured minor struts, when operating in vitro. The sounds are generated with test valves operating under a variety of conditions in a model left ventricle. It is found in the learning stage that the discriminant functions generated correctly classified almost all of the cases within the learning set. When applied to cases outside the learning set, including a recording of a clinically implanted valve, the functions correctly classify the valves. These preliminary results, for a limited number of valves, lead us to believe that the discriminant analysis of heart valve sounds is a promising noninvasive method for screening patients with implanted Bjork-Shiley convexo-concave valves.